- Boost - lists.boost.org

[rfc] Boost Graphic/Game Engine
by Felipe Magno de Almeida 12 Aug '07

12 Aug '07

Hi, There were some little interest for this in another thread here in boost. So, who has any experience in game programming, it would be interesting to layout some requirements for a new engine and complaints from other engines. See if we can find out how a game engine should really be. Regards, -- Felipe Magno de Almeida

18 35

Shared Map
by Jeremy Bruestle 12 Aug '07

12 Aug '07

Greetings, I've recently implemented a new collection type for a project of mine. It works identically to std::map, with the same algorithmic guarantees, but additionally can be 'copied' in constant time. That is: shared_map<int, int> a; for(i = 0; i < 100; i++) a[i] = i*i; // Each insert in log(n) of the map size shared_map<int, int> b = a; // This is constant time (not linear as in a std::map) b[1] = 7; // Still log(n) of course b[2] = 3; assert(a[2] == 4 && b[2] == 3 && a[6] == 36 && b[6] == 36); The trick is simple: the map is a red-black tree, but we get rid of the explicit parent pointer for each red-black node. This complicates the code, but is algorithmically uninteresting since we always descend from the root anyway and just need to track the parent as we go, either in an explicit stack or through recursion (my current implementation). The key point however is that the lack of an explicit parent allows us to use reference counting, and share children. Thus the copy from a to b simply copies the root node, and adds an additional reference count. All future changes do copy on write for shared nodes (any with a ref count of 2 or more). The does lower the overall performance for all operations somewhat, but for cases like mine where there is a need to copy large maps a lot, and each map remains quite similar to the other maps, there is a huge savings in computation and memory usage. Another nice use is: copy a map (since it's free) and iterate over it while modifying the original with no worries. One caveat, obviously multiple threads require a single global lock since even apparently different collections share internal state (or I could add a per node lock, which has the nice effect of general fine grained tree locking, but is tons of overhead, perhaps a compile time choice?). My current implementation is sufficient for my purposes, but with a little cleanup (making stronger iterator validity guarantees, better support for erasing ranges, explicit parent stack instead of recursion perhaps), it could be perhaps useful to everyone. Also, it's a little work but straightforward to generalize to shared_set, shared_multimap, etc. I was wondering if there would be any interest from people in such a library? If so, I would be willing to do the needed cleanup, if not, it suits me reasonably well as is (I don't even use erase). Also curious if anyone has seen such a thing anywhere before. It seems obvious enough to me, but perhaps it's actually new, I couldn't find anything like it when I went searching. Thanks for your time -Jeremy

12 17

Some extra files in the CVS HEAD branch
by dhruva 11 Aug '07

11 Aug '07

Hi, I was browsing the CVS repo through the web UI. I find 2 files directly under the main CVS folder (http://boost.cvs.sourceforge.net/boost/) index.html and creation_tags.hpp which are boost::interprocess. IMHO, it must be some residual files having accidently gotten into the CVS due to a bulk checkin. A request: For some of us behind the firewall, can a nightly archive of HEAD be made available? I need ASIO, INTERPROCESS and INTRUSIVE to try out a few things, this being not part of current BOOST release, there is no other way I am aware of to get those librarues. with best regards, dhruva -- Dhruva Krishnamurthy Contents reflect my personal views only!

3 3

Text encoding library.
by Atry 11 Aug '07

11 Aug '07

I wrote a text_encoding library provide generic container and iterator with charset infomation. I have used it for some weeks.Use this library, you may write code like: // convert UCS-2 to UTF-8 utf8_string us = L"some wide character: \xABCD\x1234"; cout << us << endl; string s; // move bytes to another std::string. us.swap(s); Or, use iterators: wstring ws = L"some wide character: \xABCD\x1234"; copy ( make_encoded_range<charset::utf8>(ws).begin(), make_encoded_range<charset::utf8>(ws).end(), ostreambuf_iterator<char>(std::cout) ); The version I upload to vault already have a little of tests, examples, and documents. directory url: http://boost-consulting.com/vault/index.php?direction=0&order=&directory=St… file name: text_encoding.zip

1 0

[Iterator] Adaptors for accessing keys/values in associative containers
by Joseph Gauterin 11 Aug '07

11 Aug '07

Would anyone be interested in having a couple of extra specialized iterator adaptors in Boost.Iterator? I'm thinking of 'key_iterator' and 'value_iterator' adaptors (along with convenience 'make_key_iterator' and 'make_value_iterator' functions) that allow any associative container (that follows standard library idioms) to be accessed as either a range of keys or as a range of values. They'd be used like this: std::map<std::string,int> m; std::for_each(make_value_iterator(m.begin()),make_value_iterator(m.end()),func); Where func() takes an int parameter rather than a std::pair<const std::string,int>. Using these adaptors is considerably easier than using boost.lambda for this purpose and they can be implemented to incur no (or negligable) runtime overhead when compared to iterating normally through an associative container. I find these useful - does anyone else?

6 18

[gil] trunk compile error
by Michael Marcin 11 Aug '07

11 Aug '07

Hello, From svn trunk I get a compiler error in boost\gil\gil_concept.hpp for line #1350. Inside of IteratorAdaptorConcept I get "expression must have a constant value" for BOOST_STATIC_ASSERT(is_iterator_adaptor<Iterator>()); Thanks, Michael Marcin

2 1

Re: [boost] unique sorted vector
by Ion Gaztañaga 10 Aug '07

10 Aug '07

Michael Marcin wrote: > I have a need for a low memory set. > Essentially I want an always sorted unique vector. > Is such a wrapper around vector and/or deque available in Boost? Boost.Interprocess has a sorted vector family (although implements it's own vector to make the container shared memory friendly). Interprocess offers 4 containers: flat_map, flat_multimap, flat_set and flat_multiset. You can download Interprocess from the SVN repository. You can see the code here: http://svn.boost.org/svn/boost/trunk/boost/interprocess/containers/flat_set… Having an adaptor class that converts a sequence providing random access iterators in an associative container would be a nice addition to Boost, no doubt. Regards, Ion

1 0

Compilation with Visual Studio 7.1 fails
by luca regini 10 Aug '07

10 Aug '07

While trying to compile a snapshot of the SVN repository with VS 7.1 ( on a Windows XP box ) i got server "buffer overrun" errors that resulted in about 20 targets not being compiled. MSDN says that disabling precompiled header support the problem should not manifest anymore. Is there anyone that has solved it? Thanks in advance, Luca

2 1

[svn] cleaning up old branches
by Daniel Wallin 10 Aug '07

10 Aug '07

The standard practice with subversion is to `svn delete` old branches once they have been merged. This keeps the branch directory nice and clean (and frees up the branch name for future use). Does anyone have any objections to deleting all old branches? -- Daniel Wallin Boost Consulting www.boost-consulting.com

4 5

anonymous committing?
by Vladimir Prus 10 Aug '07

10 Aug '07

Looking at http://svn.boost.org/trac/boost/changeset/38565 http://svn.boost.org/trac/boost/changeset/38564 it seems like 'anonymous' is working hard on the 'functional/hash' library. While it would be really great if zillion anonymous committers suddenly start helping out, this case seems like misconfiguration of either Subversion, or Trac. Which is this? - Volodya

2 1